Effects of Oxygen Content on Bi3Mn3O11+δ: From 45 K

Aug 13, 2010 - Effects of Oxygen Content on Bi3Mn3O11+δ: From 45 K Antiferromagnetism to Room-Temperature True Ferromagnetism. Alexei A. Belik* and ...
3 downloads 0 Views 2MB Size
Download PDF
Published on Web 08/13/2010

Effects of Oxygen Content on Bi3Mn3O11+δ: From 45 K Antiferromagnetism to Room-Temperature True Ferromagnetism Alexei A. Belik* and Eiji Takayama-Muromachi International Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan Received May 19, 2010; E-mail: [email protected]

Abstract: The effects of oxygen content on the structural, physical, and chemical properties of Bi3Mn3O11 with KSbO3-type structure have been investigated. It was found that the oxygen content in Bi3Mn3O11+δ can vary over a wide δ range, keeping the same cubic structure (space group Pn3j , a ) 9.12172(5) Å for δ ) -0.5, a ) 9.13784(8) Å for δ ) 0, and a ) 9.09863(7) Å for δ ) 0.6) and semiconducting properties of the material. At the same time, magnetic properties change from true antiferromagnetic with TN ) 45 K for δ ) -0.5 to true ferromagnetic with TC ) 307 K for δ ) +0.6. Bi3Mn3O11 (δ ) 0) shows ferrimagneticlike properties with TC ) 150 K and features typical for a re-entrant spin-glass below 30 K. Noticeable changes of the magnetic transition temperature and magnetism in Bi3Mn3O11+δ with δ can be compared with changes of the magnetic and electronic properties of LaMnO3+δ, BiMnO3+δ, high-temperature copper superconductors (e.g., YBa2Cu3O7+δ), and other cuprates. Bi3Mn3O11.6 shows a new record high TC among insulating/semiconducting true ferromagnets. Our results demonstrate that the oxygen content can vary for the same cation composition in KSbO3-type materials, and the oxygen content can be increased up to BiMnO3.867 (Bi3Mn3O11.6).

1. Introduction

ABO3 compounds crystallize in a number of structure types, including perovskite, pyroxene, corundum, ilmenite (ordered corundum), hexagonal manganites, bixbyite, rare earth sesquioxide structures, PbReO3, KSbO3, AlFeO3, and others.1 The KSbO3-type family (space group Pn3j) has only about a dozen of representatives, e.g., KSbO3, KIrO3,2 Bi3GaSb2O11,3,4 Bi3AlSb2O11,4 Bi2NaSb3O11,5 Bi3M3O11 (M ) Ru,6-8 Re,9 Os8), La3Ru3O11,10-12 and Pb6Re6O19.13 Nevertheless, this class of materials is quite interesting because the oxygen content changes from ABO3 to ABO3.667. The KSbO3-type structure is built from (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

Giaquinta, D. M.; zur Loye, H.-C. Chem. Mater. 1994, 6, 365. Hoppe, R.; Claes, K. J. Less-Common Metals 1975, 43, 129. Sleight, W.; Bouchard, R. J. Inorg. Chem. 1973, 12, 2314. (a) Ismunandar; Kennedy, B. J.; Hunter, B. A. Solid State Commun. 1998, 108, 649. (b) Ismunandar; Kennedy, B. J.; Hunter, B. A. J. Solid State Chem. 1996, 127, 178. Champarnaud-Mesjard, J. C.; Frit, B.; Aftati, A.; El Farissi, M. Eur. J. Solid State Inorg. Chem. 1995, 32, 495. (a) Facer, G. R.; Elcombe, M. M.; Kennedy, B. J. Aust. J. Chem. 1993, 46, 1897. (b) Abraham, F.; Thomas, D. Bull. Soc. Chim. Fr. 1975, 25. Tsuchida, K.; Kato, C.; Fujita, T.; Kobayashi, Y.; Sato, M. J. Phys. Soc. Jpn. 2004, 73, 698. Fujita, T.; Tsuchida, K.; Yasui, Y.; Kobayashi, Y.; Sato, M. Physica B 2003, 329, 743. Suzuki, H.; Ozawa, H.; Sato, H. J. Phys. Soc. Jpn. 2007, 76, 044805. (a) Cotton, F. A.; Rice, C. E. J. Solid State Chem. 1978, 25, 137. (b) Abraham, F.; Trehoux, J.; Thomas, D. Mater. Res. Bull. 1978, 13, 805. Khalifah, P.; Nelson, K. D.; Jin, R.; Mao, Z. Q.; Liu, Y.; Huang, Q.; Gao, X. P. A.; Ramirez, A. P.; Cava, R. J. Nature 2001, 411, 669. Khalifah, P.; Cava, R. J. Phys. ReV. B 2001, 64, 085111. Abakumov, A. M.; Shpanchenko, R. V.; Antipov, E. V. Z. Anorg. Allg. Chem. 1998, 624, 750.

12426

9

J. AM. CHEM. SOC. 2010, 132, 12426–12432

the corner- and edge-shared SbO6 octahedra forming a threedimensional framework with channels.3 The channels are filled with K+ ions in KSbO3 (K · SbO3) and with Bi3O2 units in Bi3Ru3O11 (Bi3O2 · 3RuO3) in an ordered manner. Disordered filling of the channels by K+ ions in KBiO3,14 Na+ ions in NaSbO3,15 and Sr2+ ions in Sr2Re3O99 resulted in space group Im3j. The channels can also be filled by other units, e.g. La4O, forming La4Ru6O19 (La4O · 6RuO3) and related compounds La4M6O199 that deviate from the A:B ) 1:1 stoichiometry and have a different space group of I23.11,12 Variations in the oxygen content for the same cation composition have not been reported for the KSbO3-type materials, even though some of them have 4d and 5d transition metals with variable oxidation states. The oxygen content is known to have crucial and dramatic effects on the properties of materials, e.g., on magnetic and electronic properties of high-temperature copper superconductors (e.g., YBa2Cu3O7+δ)16 orperovskites(e.g.,LaMnO3+δ,17,18 BiMnO3+δ,19-21 (14) Nguyen, T. N.; Giaquinta, D. M.; Davis, W. M.; zur Loye, H. C. Chem. Mater. 1993, 5, 1273. (15) Hong, H. Y.-P.; Kafalas, J. A.; Goodenough, J. B. J. Solid State Chem. 1974, 9, 345. (16) Strobel, P.; Capponi, J. J.; Chaillout, C.; Marezio, M.; Tholence, J. L. Nature 1987, 327, 306. (17) Topfer, J.; Goodenough, J. B. J. Solid State Chem. 1997, 130, 117. (18) Maurin, I.; Barboux, P.; Lassailly, Y.; Boilot, J. P.; Villain, F.; Dordor, P. J. Solid State Chem. 2001, 160, 123. (19) Sundaresan, A.; Mangalam, R. V. K.; Iyo, A.; Tanaka, Y.; Rao, C. N. R. J. Mater. Chem. 2008, 18, 2191. (20) Belik, A. A.; Kolodiazhnyi, T.; Kosuda, K.; Takayama-Muromachi, E. J. Mater. Chem. 2009, 19, 1593. (21) Belik, A. A.; Kodama, K.; Igawa, N.; Shamoto, S.; Kosuda, K.; Takayama-Muromachi, E. J. Am. Chem. Soc. 2010, 132, 8137. 10.1021/ja1043598  2010 American Chemical Society

Effects of Oxygen Content on Properties of Bi3Mn3O11+δ

ARTICLES

and (BiMn3)Mn4O12).22 In LaMnO3+δ,17,18 the change in δ results in changes from an antiferromagnetic insulator to a ferromagnetic insulator to a ferromagnetic metal; at the same time, crystal symmetries are also changed. Recently, we have succeeded in preparing the first example of a KSbO3-type material containing a 3d transition metal, namely, Bi3Mn3O11.23 Bi3Mn3O11 shows quite interesting magnetic properties: a ferrimagnetic-like transition near 150 K with features typical for re-entrant spin-glass below 30 K. We have also demonstrated the possibility of forming oxygen-deficient Bi3Mn3O11-δ samples by heating Bi3Mn3O11.23 The properties of Bi3Mn3O11-δ and Bi3Mn3O11+δ have not been determined yet. In this work, we demonstrate for the first time that the oxygen content of the KSbO3-type materials can be increased further up to ABO3.867 in Bi3Mn3O11.6 (Bi3O2.6 · 3MnO3), meaning that there is additional space in the channels of the structure. We also show that the oxygen content can vary for the same cation composition in KSbO3-type materials. A wide variation of δ in Bi3Mn3O11(δ keeps the same cubic structure and semiconducting properties of the material, compared, for example, with perovskites. At the same time, magnetic properties change from antiferromagnetic with TN ) 45 K for δ ) -0.5 to true ferromagnetic with TC ) 307 K for δ ) +0.6. Bi3Mn3O11.6 shows a new record high ferromagnetic Curie temperature (TC) among insulating/semiconducting true ferromagnets.24 2. Experimental Section 2.1. Synthesis and Determination of the Composition. Bi3Mn3O11 was prepared from stoichiometric mixtures of Bi2O3 (99.9999%, Rare Metallic Co. Ltd.), MnO2 (99.997%, Alfa Aesar), and Bi2O4.20 (99%, High Purity Chemicals Ltd.) according to the following reaction:23

13/12Bi2O3 + 3MnO2 + 5/12Bi2O4.20 f Bi3Mn3O11 (1) The phase purity and oxygen content of MnO2 were confirmed by X-ray powder diffraction (XRD) and thermogravimetric analysis (TGA). MnO2 was the well-crystallized single-phase R modification. MnO2 was heated in air to 923 K over 5 h and soaked there for 24 h (the experiment was performed in a conventional muffle furnace; the sample weights were measured before and after annealing). The final product was well-crystallized single-phase Mn2O3. The oxygen content calculated from the weight loss corresponded to MnO2.01(1). The oxygen content of a commercial Bi2O5 was found to be Bi2O4.20(2) by TGA. The commercial Bi2O5 was heated in an SII Exstar 6000 (TG-DTA 6200) system to 770 K (10 K/min, Pt holder) in air, and the weight loss gave the composition of Bi2O4.20(2). The final product after the TGA heat treatment was well-crystallized single-phase Bi2O3. The synthesis of Bi3Mn3O11 according to reaction 1 was performed in a belt-type high-pressure apparatus at 6 GPa and 1600 K for 40 min in Pt capsules.23 After heat treatment, the samples were quenched to room temperature, and the pressure was slowly released. The resultant samples were dense black pellets. No noticeable change in the weight of the Pt capsules was found after the synthesis, indicating that the target oxygen content did not change during the reaction. (22) (a) Imamura, N.; Karppinen, M.; Motohashi, T.; Fu, D.; Itoh, M.; Yamauchi, H. J. Am. Chem. Soc. 2008, 130, 14948. (b) Imamura, N.; Karppinen, M.; Yamauchi, H. Chem. Mater. 2009, 21, 2179. (23) Belik, A. A.; Takayama-Muromachi, E. J. Am. Chem. Soc. 2009, 131, 9504. (24) Rogado, N. S.; Li, J.; Sleight, A. W.; Subramanian, M. A. AdV. Mater. 2005, 17, 2225.

Figure 1. Thermogravimetric curves of Bi3Mn3O11 (two different samples,

blue and green curves) and Bi3Mn3O11.6 (black curve) in high-purity Ar at a heating rate of 5 K/min. Note that the weight loss is given in δ for Bi3Mn3O11+δ and Bi3Mn3O11.6+δ.

Bi3Mn3O11.6 was prepared using the same synthesis conditions from stoichiometric mixtures of Bi2O3, MnO2, and Bi2O4.20, taken for the target composition of Bi3Mn3O12 according to the following reaction:

0.25Bi2O3 + 3MnO2 + 1.25Bi2O4.20 f Bi3Mn3O11.6 + 0.2O2 (2) The weight of Pt capsules after the synthesis changed noticeably, indicating that the Pt capsules were not able to maintain the generated oxygen pressure, and some oxygen escaped from the Pt capsules. The oxygen content of the final products was determined by TGA in a mixture of 3% H2 + 97% Ar using a Perkin-Elmer Pyris 1 TGA system in Al2O3 holders (the samples were heated to 873 K at a heating rate of 5 K/min and soaked there for 1 h). The samples decomposed to mixtures of Bi and MnO. The calculated oxygen content was Bi3Mn3O11.00(1) and Bi3Mn3O11.60(1) (see the Supporting Information). We note that the oxygen content determination by TGA or other chemical methods is based on certain assumptions, as discussed in ref 20. If some assumptions are not kept, a systematic shift may result. For example, the presence of impurities will affect the results of both TGA and (iodometric) titration techniques. The Bi3Mn3O10.5 sample (used for the property measurements) was prepared from Bi3Mn3O11.60. Bi3Mn3O11.60 was heated to 750 K and cooled to room temperature inside an oven in a magnetometer (Quantum Design MPMS). The average heating/cooling rate was about 1.5 K/min. Inside the oven, vacuum conditions were kept (about 1.3 kPa). Bi3Mn3O∼10.5 can also be prepared from Bi3Mn3O11.0 and Bi3Mn3O11.6 by heating to 750-770 K in highpurity Ar (at a heating rate of 5 K/min and soaking time of 1 h) (Figure 1). The cation ratio of Bi3Mn3O11 and BiMnO3 (used for comparison) was determined by electron probe microanalysis (EPMA) using a JEOL JXA-8500F instrument. The surface was polished on a fine alumina (1 µm)-coated film before the EPMA measurements. MnO and Bi4Ge3O12 were used as standard materials for Mn and Bi. The Mn:Bi ratio determined by EPMA was 1.05(2) in Bi3Mn3O11 and 1.05(6) in BiMnO3. 2.2. Physical Properties. XRD data were collected at room temperature on a RIGAKU Ultima III diffractometer using Cu KR radiation (2θ range of 10-120°, step width of 0.02°, and counting time of 10 s/step). Synchrotron XRD data for Bi3Mn3O10.50 and Bi3Mn3O11.60 were collected at room temperature using a large J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010 12427

ARTICLES

Belik and Takayama-Muromachi

Figure 2. X-ray powder diffraction patterns (measured with Cu KR radiation at room temperature) of Bi3Mn3O10.5, Bi3Mn3O11.6, and Bi3Mn3O11. The refined lattice parameters are given. Possible Bragg reflections (space group Pn3j) are indicated by tic marks for Bi3Mn3O11.

Debye-Scherrer camera at the BL02B2 beamline25 of SPring-8. The incident beam from a bending magnet was monochromatized to λ ) 0.41916 Å. The samples were contained in (boro)glass capillary tubes with an inner diameter of 0.2 mm, and the capillary tubes were rotated during measurements. The synchrotron XRD data were collected in a 2θ range from 1° to 75° with a step interval of 0.01° (the data from 2.5° to 52.5° were used in the refinements). Laboratory XRD and synchrotron XRD data were analyzed by the Rietveld method with RIETAN-2000 software.26 High-temperature synchrotron XRD data of Bi3Mn3O11 were collected at the BL02B2 beamline with λ ) 0.77412 Å. Magnetic susceptibilities, χ ) M/H, of Bi3Mn3O11(δ were measured on a SQUID magnetometer (Quantum Design, MPMS) between 2 and 400 K in applied fields of 100 Oe, 1 kOe, and 10 kOe under both zero-field-cooled (ZFC) and field-cooled (FC; on cooling) conditions. The measurements between 300 and 750 K were performed using an oven attachment on heating and cooling at 10 kOe. Isothermal magnetization measurements were performed at 5, 100, 200, 300, and 350 K between -50 and 50 kOe and between 270 and 350 K with a step of 2 K from 50 to 0 kOe for Bi3Mn3O11.6. TC and TN were roughly defined by the peak positions on the 100 Oe dχ/dT vs T and d(χT)/dT vs T curves, respectively. The specific heat, Cp, at magnetic fields of 0 and 70 kOe was recorded between 2 and 300 K (using an N Apiezon grease to make good thermal contacts between sample and holder) and between 250 and 400 K (using an H Apiezon grease) on cooling by a pulse relaxation method using a commercial calorimeter (Quantum Design PPMS). Direct current (dc) electrical resistivity was measured between 10 and 400 K by the conventional four-probe method using a Quantum Design PPMS with the dc-gage current of 2 mA. The dc-gage current was reduced automatically at low temperatures to allow the resistivity measurements. Resistivity became too high to be measured with our system below 300 K for Bi3Mn3O10.5, 130 K for Bi3Mn3O11, and 180 K for Bi3Mn3O11.6. At 300 K, Bi3Mn3O11.6 showed an almost linear increase of resistivity as a function of magnetic field, reaching about a 15% increase at 90 kOe (see the Supporting Information). Pellets with approximate sizes of 5 × 2 × 1 mm3 were used for the measurements. Dielectric measurements (25) Nishibori, E.; Takata, M.; Kato, K.; Sakata, M.; Kubota, Y.; Aoyagi, S.; Kuroiwa, Y.; Yamakata, M.; Ikeda, N. Nucl. Instrum. Methods Phys. Res. Sect. A 2001, 467-468, 1045. (26) Izumi, F.; Ikeda, T. Mater. Sci. Forum 2000, 321-324, 198. 12428

J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010

of Bi3Mn3O10.5 were performed between 5 and 300 K in the frequency range from 100 Hz to 1 MHz; no anomalies were detected near TN (see the Supporting Information). 3. Results 3.1. X-ray Powder Diffraction of Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6. Figure 2 depicts XRD patterns of Bi3Mn3O10.5,

Bi3Mn3O11, and Bi3Mn3O11.6. All the samples were almost single-phase with traces of unidentified impurities. Bi3Mn3O11(δ crystallize in the KSbO3-type structure (space group Pn3j) with a ) 9.12172(5) Å (δ ) -0.5), 9.13784(8) Å (δ ) 0), and 9.09863(7) Å (δ ) 0.6). A non-monotonic change of the lattice parameter with δ is probably the combination of different effects: contraction of the unit cell due to the increased amount of Mn5+ and expansion of the unit cell due to the increased number of oxygen ions in the channels. A non-monotonic change of the lattice parameter was also observed during in situ high-temperature synchrotron XRD measurements,23 where the a parameter first jumps in Bi3Mn3O10.9 compared with Bi3Mn3O11 and then drops in Bi3Mn3O10.5 (see also the Supporting Information). 3.2. Crystal Structure Analysis of Bi3Mn3O10.5 and Bi3Mn3O11.6.

Fractional coordinates of Bi3Mn3O1123 were used as initial ones in the crystal structure analysis of Bi3Mn3O10.5 and Bi3Mn3O11.6. Fractional coordinates of the Bi1 site in Bi3Mn3O11.6 were refined using a constraint z ) y (similar to Bi3Mn3O11),23 even though the Bi1 site is located in a general position (x,y,z). This constraint was used because the y and z coordinates were the same within standard deviations, and estimated standard deviations of y and z were about 1 order of magnitude larger when the y and z coordinates were refined independently. On the other hand, the x, y, and z coordinates of the Bi1 site could be refined in Bi3Mn3O10.5. Attempts to split the Bi2 atom from the 4b site to the 8e site were unsuccessful in Bi3Mn3O11 and Bi3Mn3O11.6 (the x coordinate was almost zero, and the estimated standard deviation was huge). On the other hand, the Bi2 site was successfully split in Bi3Mn3O10.5. With an occupation factor (g) of 1 for the O1, O2, and O3 sites, the thermal parameters (B) were normal in Bi3Mn3O11.6, while B(O1) converged to 8.5(6) Å2 in Bi3Mn3O10.5. This fact shows that the oxygen vacancies

Effects of Oxygen Content on Properties of Bi3Mn3O11+δ

ARTICLES

Table 1. Structure Parameters of Bi3Mn3O10.5 and Bi3Mn3O11.6 at

Room Temperaturea site

Wyckoff position

g

x

y

Bi1 Bi2 Mn O1 O2 O3

24h 8e 12g 8e 12f 24h

1/3 1/2 1 0.75 1 1

Bi3Mn3O10.5 0.4164(3) 0.3879(6) 0.0091(2) )x 0.4041(3) 0.75 0.1456(8) )x 0.6163(11) 0.25 0.5841(8) 0.2500(7)

Bi1 Bi2 Mn O1 O2 O3

24h 4b 12g 8e 12f 24h

1/3 1 1 1 1 1

Bi3Mn3O11.6 0.4009(2) 0.38175(13) 0 0 0.4044(2) 0.75 0.1455(4) )x 0.6147(8) 0.25 0.5908(6) 0.2445(5)

z

B (Å2)

0.3736(5) )x 0.25 )x 0.25 0.5435(11)

1.62(5) 0.84(3) 0.30(4) 1.0(4) 0.9(2) 1.2(2)

)y 0 0.25 )x 0.25 0.5403(8)

0.61(2) 0.730(11) 0.46(3) 0.4(2) 0.67(14) 0.62(13)

a Space group is Pn3j (No. 201) at origin choice 2, Z ) 4. g is the occupation factor. Bi3Mn3O10.5: a ) 9.11919(8) Å and V ) 758.348(11) Å3, Rwp ) 4.46%, Rp ) 3.22%, RB ) 6.99%, and RF ) 7.95%. d(Mn-O2) ) 1.860(6) Å × 2, d(Mn-O3) ) 1.886(10) Å × 2, d(Mn-O3) ) 1.976(10) Å × 2. Bi3Mn3O11.6: a ) 9.11151(6) Å and V ) 756.434(9) Å3, Rwp ) 3.44%, Rp ) 2.28%, RB ) 1.96%, and RF ) 1.81%. d(Mn-O2) ) 1.871(4) Å × 2, d(Mn-O3) ) 1.908(6) Å × 2, d(Mn-O3) ) 1.912(7) Å × 2.

are located at the O1 site in Bi3Mn3O10.5. The same result was obtained during the structural analysis of Bi3Mn3O11 using in situ high-temperature synchrotron XRD data at 740 K, where the oxygen content was close to Bi3Mn3O10.5 (see the Supporting Information of ref 23). Using synchrotron XRD data, we could not localize additional oxygen atoms in Bi3Mn3O11.6. Final structural parameters, R indexes, and Mn-O bond lengths of Bi3Mn3O10.5 and Bi3Mn3O11.6 at room temperature are listed in Table 1. Figure 3 shows fragments of the crystal structure of Bi3Mn3O10.5. The Supporting Information gives other bond lengths and observed, calculated, and difference synchrotron XRD patterns.

Figure 3. Fragments of the crystal structure of Bi3Mn3O10.5 with the split Bi1 and Bi2 atoms. The MnO6 octahedra are shown in gray. The O1 atoms in the channels are given by black circles. (a) The view along the channel (111) direction. (b) The view along the a axis.

3.3. Magnetic Properties of Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6. Figure 4 shows the inverse magnetic susceptibilities

as a function of temperature (χ-1 vs T) for three samples and actually one of the preparation routes for Bi3Mn3O10.5. The curves were fit by the simple Curie-Weiss equation (χ ) µeff2/ [8(T - θ)]) between 400 and 550 K. The effective magnetic moment (µeff) per formula unit (f.u.) and the Curie-Weiss temperature (θ) were calculated as follows: µeff ) 5.84 µB and θ ) +217 K in Bi3Mn3O11 (the expected magnetic moment, µexp, is 6.16 µB); µeff ) 5.37 µB and θ ) +337 K in Bi3Mn3O11.6 (µexp ) 5.57 µB); and µeff ) 6.58 µB and θ ) -80 K in Bi3Mn3O10.5 (µexp ) 6.70 µB). The changes in the experimental µeff values are in very good agreement with the changes of the oxidations states of Mn atoms (Bi3Mn4+3O10.5, Bi3Mn4+2Mn5+O11, and, for simplicity, Bi3Mn4+Mn5+2O11.5). This fact also shows that Mn ions are responsible for the charge compensation in Bi3Mn3O11(δ, not Bi ions. Bi3Mn3O11.6 demonstrates clearly a ferromagnetic-like transition with TC ) 310-315 K (a temperature step during the magnetization measurements was 5 K in this temperature range), and it has a large positive Curie-Weiss temperature. (TC will be estimated more precisely below.) Bi3Mn3O10.5 shows clearly an antiferromagnetic transition with the Neel temperature, TN, of 45 K, and it has a negative Curie-Weiss temperature. The ZFC and FC curves of Bi3Mn3O11.6 were almost constant below about 250 K, without any additional anomalies at low temperatures (see the Supporting Information) compared with Bi3Mn3O11, where features typical for a re-entrant spin-glass were observed below 30 K.23 The

Figure 4. Inverse magnetic susceptibilities of Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6. The measurements were performed at 10 kOe between 2 and 400 K (filled symbols) and between 300 and 750 K (open symbols) using an oven attachment. The linear curves between 400 and 550 K depict the Curie-Weiss fits, with the fitting parameters given in the figure. The arrows demonstrate the preparation route of Bi3Mn3O10.5 from Bi3Mn3O11.6 inside the oven during the measurements.

magnetic properties of Bi3Mn3O11 are consistent with ferrimagnetic behavior.23 Bi3Mn3O11 is called a ferrimagnet because its saturated magnetization reaches only a portion (3.19 µB at 5 K and 50 kOe)23 of the maximum value (8 µB), even though it has a positive Curie-Weiss temperature of 217 K. The Curie-Weiss temperature is actually a sum of different interactions between magnetic ions. In most cases, a positive Curie-Weiss temperature corresponds to ferromagnets, and a J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010 12429

ARTICLES

Belik and Takayama-Muromachi

Figure 5. Specific heat data for Bi3Mn3O10.5 and Bi3Mn3O11.6 (inset) at 0

Oe (open symbols) and 70 kOe (filled symbols) plotted as Cp/T vs T. An anomaly near 350 K is due to a sample holder.

Figure 6. Isothermal magnetization curves for Bi3Mn3O11.6 at 5, 100, 200, 300, and 350 K (circles) and for Bi3Mn3O10.5 at 5 K (triangles). Inset shows pieces of Bi3Mn3O11.6 attached to magnetized tweezers at room temperature.

negative Curie-Weiss temperature corresponds to antiferromagnets. However, there are exceptions. For example, LaMnO3 is a classical (canted) antiferromagnet. However, it has a positive Curie-Weiss temperature of 52 K because the intraplane ferromagnetic interactions are much stronger than the interplane antiferromagnetic interactions.27 The positive CurieWeiss temperature of Bi3Mn3O11 reflects the fact that ferromagnetic-type interactions are dominant. It is also difficult to consider Bi3Mn3O11 as a canted antiferromagnet because of its positive Curie-Weiss temperature and large saturated magnetization. Figure 5 depicts specific heat data of Bi3Mn3O10.5 and Bi3Mn3O11.6. Bi3Mn3O10.5 shows a strong anomaly near TN, which is almost independent of a magnetic field. This behavior is typical for antiferromagnets, and the data unambiguously confirmed the onset of the long-range antiferromagnetic order in Bi3Mn3O10.5. Bi3Mn3O11.6 has a weak anomaly near TC, and a magnetic field smears the transition to a high temperature region, which is typical for ferromagnets. The specific heat anomaly in Bi3Mn3O11.6 confirms the onset of long-range ferromagnetic order. The magnitude of an anomaly near TC or TN (or the amount of the released entropy) depends on many factors. The higher the temperature, the weaker the magnetic anomalies because of an increased contribution from the lattice. Competition between different interactions or frustration and structural disorder also significantly suppress magnetic anomalies near TC or TN. The above reasons may be responsible for the weak magnetic anomaly at TC in the specific heat of Bi3Mn3O11.6. Very weak magnetic anomalies at TC were also found in the specific heat of LaMnO3+δ.28 We note that no specific heat anomalies were observed in Bi3Mn3O11.23 Isothermal magnetization curves (M vs H) are shown in Figure 6. In Bi3Mn3O11.6 at 50 kOe, the magnetization reaches 7.33 µB/f.u. at 5 K (the value being very close to the full magnetization) and 3.51 µB/f.u. at 300 K. These results show that Bi3Mn3O11.6 is a true ferromagnet. Bi3Mn3O11.6 behaves as a very soft ferromagnet with negligible remnant magnetization (0.08 µB/f.u. at 5 K) and coercive field similar to those of ferrimagnetic metal Sr2FeMoO629 and ferromagnetic semiconductor BiMnO3.30 Bi3Mn3O11.6 is attached to permanent magnets

and magnetized objects at room temperature (see the inset of Figure 6). At 350 K, an almost linear behavior was observed on the M vs H curves of Bi3Mn3O11.6. In antiferromagnetic Bi3Mn3O10.5, linear M vs H curves were found at 5 K without any signs of hysteretic behavior.

(27) Zhou, J.-S.; Goodenough, J. B. Phys. ReV. B 1999, 60, 15002. (28) Ghivelder, L.; Castillo, I. A.; Gusmao, M. A.; Alonso, J. A.; Cohen, L. F. Phys. ReV. B 1999, 60, 12184. (29) Kobayashi, K. L.; Kimura, T.; Sawada, H.; Terakura, K.; Tokura, Y. Nature 1998, 395, 677.

To estimate TC of Bi3Mn3O11.6 more precisely, we used the Arrott plots (M2 vs H/M; Figure 7a). The Arrott plots deviate from linear behavior near TC, similar to some other ferromagnets.31 Therefore, we estimated the inverse initial susceptibilities (χ0-1) by smooth extrapolation of the M2 vs H/M curves above

12430

J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010

Figure 7. (a) Arrott plots (M2 vs H/M) for Bi3Mn3O11.6 from 270 to 350 K with a step of 2 K. (b) Inverse initial susceptibilities (χ0-1; circles) versus temperature for Bi3Mn3O11.6. The line shows the fitting results, with the equation and fitting parameters given in the figure.

Effects of Oxygen Content on Properties of Bi3Mn3O11+δ

ARTICLES

Even though the target composition was Bi3Mn3O12, the resultant sample had the composition of Bi3Mn3O11.6. This fact shows that 11.6 is the maximum oxygen content at the synthesis conditions, and the structure cannot accommodate more oxygen. The TGA data in hydrogen (see the Supporting Information) and Ar (Figure 1) clearly confirmed that Bi3Mn3O11.6 contains more oxygen than Bi3Mn3O11. We note that Bi3Mn3O10.5 could only be prepared from Bi3Mn3O11 or Bi3Mn3O11.6 by heating in an inert atmosphere or in a vacuum. The direct high-pressure high-temperature synthesis of Bi3Mn3O10.5 produced a sample containing a lot of impurities (see the Supporting Information). We should also emphasize step-like weight changes in Bi3Mn3O11 and Bi3Mn3O11.6 on heating (Figure 1). This fact shows that other compositions with intermediate oxygen contents can be stabilized, for example, Bi3Mn3O10.9. Oxygen content has a drastic effect on the magnetism of Bi3Mn3O11(δ and on the nature of magnetic interactions. Bi3Mn3O10.5, with the transition metal in one oxidation state of 4+, has antiferromagnetic interactions between Mn4+ ions and a rather low TN of 45 K. Note that Bi3Mn3O10.5 behaves as a true antiferromagnet, that is, without any signs of spin canting. Introduction of Mn5+ ions into the system seems to produce

ferromagnetic-type interactions and increase the strength of interactions and, therefore, transition temperatures. Bi3Mn3O11, with one-third of Mn5+ ions, shows ferrimagnetic-like properties and a transition temperature of about 150 K.23 The appearance of spin-glass properties below 30 K in Bi3Mn3O11 gives support for strong competition between antiferromagnetic and ferromagnetic interactions. Bi3Mn3O11.6, with about two-thirds of Mn5+, shows room-temperature ferromagnetism. Noticeable changes of the magnetic transition temperature and magnetism in Bi3Mn3O11+δ with δ can be compared with significant changes of the magnetic and electronic properties of perovskite-like LaMnO3+δ17,18 and BiMnO3+δ,19–21 high-temperature copper superconductors (e.g., YBa2Cu3O7+δ),16 and other cuprates (e.g., Sr2Cu(Re0.69Ca0.31)O6+δ).32 However, compared with perovskitetype materials and cuprates, KSbO3-type materials have been investigated poorly, especially from the theoretical point of view.12 In perovskites LaMnO3+δ17,18 and BiMnO3+δ,19–21 the crystal symmetry changes with changing δ, in comparison with Bi3Mn3O11(δ, where the cubic structure is stable over a wide δ range. The structural stability of Bi3Mn3O11(δ can probably be explained by its channel-like feature (Figure 3): (additional) oxygen atoms should be located in the channels, and they are removed from the channels as confirmed by the structural analysis of Bi3Mn3O10.5. The formula Bi3Mn3O11(δ seems to reflect the real changes in the composition. In perovskites, cation vacancies La1-xMn1-xO3 are actually formed, even though the formula is written as LaMnO3+δ for simplicity. Detailed structural investigations and localization of additional oxygen in Bi3Mn3O11.6 are left for future work. Some structural differences among Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6 can be seen from the available data. The formation of oxygen vacancies in the channels of the Bi3Mn3O10.5 structure results in increased disordering of Bi atoms in the channels: the Bi1 atoms are split by about 0.5 Å in Bi3Mn3O10.5, 0.35 Å in Bi3Mn3O11, and 0.25 Å in Bi3Mn3O11.6; the Bi2 atoms are split by about 0.3 Å in Bi3Mn3O10.5 and show no splitting in Bi3Mn3O11 and Bi3Mn3O11.6. The increased disordering of Bi atoms can be explained by the fact that Bi atoms become underbonded with the formation of oxygen vacancies. We note that RB and RF indexes in Bi3Mn3O10.5 were significantly larger that those of Bi3Mn3O1123 and Bi3Mn3O11.6 (Table 1). This fact may indicate additional structural disorder, which cannot be represented by the split-atom model. The origin of ferromagnetism in LaMnO3-based materials has been investigated extensively using experimental and theoretical approaches (the parent compound LaMnO3 is an antiferromagnetic insulator). The double-exchange mechanism between Mn3+ and Mn4+ is capable of describing many experimental data on these materials in the ferromagnetic metallic regime.33 However, there exists a ferromagnetic semiconducting phase in LaMnO3+δ,17,18 La1-xCaxMnO3,33 and La1-xSrxMnO3 in very narrow δ and x ranges. For example, in LaMnO3+δ, semiconducting ferromagnetic phases with TC ≈ 200 K exist near δ ≈ 0.10. Some analogies can be seen in Bi3Mn3O11(δ: Bi3Mn3O10.5 with Mn4+ ions is an antiferromagnet, and Bi3Mn3O11.6 is a semiconducting ferromagnet containing Mn4+ and Mn5+ ions. If a metallic ferromagnetic phase exists in Bi3Mn3O11+δ, it should appear for δ > 0.6, which could not be achieved in our experiments. Magnetic and transport properties of LaMnO3based materials strongly depend on Mn-O-Mn bond angles,

(30) Kimura, T.; Kawamoto, S.; Yamada, I.; Azuma, M.; Takano, M.; Tokura, Y. Phys. ReV. B 2003, 67, 180401. (31) Kouvel, J. S.; Fisher, M. E. Phys. ReV. 1964, A136, 1626.

(32) Isobe, M.; Kimoto, K.; Takayama-Muromachi, E. J. Magn. Magn. Mater. 2007, 312, 91. (33) Edwards, D. M. AdV. Phys. 2002, 51, 1259.

Figure 8. Direct current resistivity data for Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6. The data are plotted as F (in logarithmic scale) vs 103/T. The bold lines show temperature regions used for the estimation of the activation energy.

TC on the H/M axis. The χ0-1 vs T curve (Figure 7b) was fit by the simple equation,31 χ0-1 ) A(T/TC - 1)γ, with the parameters of TC ) 307.0(2) K and γ ) 1.292(11). 3.4. Transport Properties of Bi3Mn3O10.5, Bi3Mn3O11, and Bi3Mn3O11.6. Figure 8 demonstrates that Bi3Mn3O10.5, Bi3Mn3O11,

and Bi3Mn3O11.6 have semiconducting-type conductivity. The activation energies in Bi3Mn3O11 (between 225 and 400 K) and Bi3Mn3O11.6 (between 180 and 230 K) were almost the same (0.18 eV). This fact shows that these compounds behave as narrow-band semiconductors. Bi3Mn3O11.6 shows a small change of the activation energy above TC. The activation energy in Bi3Mn3O10.5 was about 0.31 eV. Bi3Mn3O11.6 should have additional oxygen atoms in the channel of the structure. Mobility of oxygen atoms at higher temperatures may be responsible for the change of the activation energy. 4. Discussion

J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010 12431

ARTICLES

Belik and Takayama-Muromachi

which are 180° in the ideal perovskite structure with cornershared MnO6 octahedra. Bi3Mn3O11 has a different framework made of edge- and corner-shared MnO6 octahedra (Figure 3). The Mn-O-Mn bond angles are about 100° and 130° in Bi3Mn3O11.23 Therefore, it is difficult to apply classical rules with 90° or 180° cation-anion-cation bridges (e.g., Goodenough-Kanamori rules). It is interesting to note that Bi3Mn3O11(δ maintains semiconducting properties, while similar compounds Bi3M3O11 with 4d and 5d transition metals have been reported to have metallic-like conductivity.7–9,11 Materials possessing ferromagnetic properties at room temperature are of great practical interest.34-38 They are used in magnetic data storage, transformer cores, permanent magnets, and so forth. Ferromagnetic dielectrics for spin polarization34 and ferromagnetic diluted semiconductors35,36 for spin electronics have recently attracted much attention. Ferromagnetic-like properties are observed in true ferromagnets (that is, where the main interaction between magnetic ions is ferromagnetic), in canted antiferromagnets or so-called weak ferromagnets, and in ferrimagnets (where the main interaction between magnetic ions is antiferromagnetic, and ferromagnetic-like properties appear due to spin canting or uncompensated magnetic sublattices).38 True ferromagnets usually have metallic conductivity, for example, the elemental metals Fe (with magnetic transition temperature TC ) 1043 K), Ni (TC ) 627 K),31 and Co (TC ) 1388 K)38 and oxides CrO2 (TC ) 386 K),38 SrRuO3 (TC ) 160 K),39 LaMnO3+δ (δ ≈ 0.14, TC ≈ 200 K),17,18 La1-xSrxMnO3 (x ≈ 0.3-0.5, TC ≈ 350 K),33 and EuO (Eu-rich, TC ) 69 K).38,40 Antiferromagnets (including canted antiferromagnets and ferrimagnets) are usually insulators or semiconductors. Insulating/semiconducting ferromagnets are always exceptions to the general rule and, therefore, of interest not only from the practical point of view but also from the viewpoint of the mechanism. Many ferrimagnets and canted antiferromagnets have high magnetic ordering temperatures well above room temperature, for example, magnetite Fe3O4 (TC ) 858 K),38 known since ancient times, and perovskites Sr2FeMoO6 (TC ≈ 450 K),29,37 Sr2Cu(Re0.69Ca0.31)O6 (TC ) 440 K),32 BiCu3Mn4O12 (TC ) 350 K),41 and Sr1-xYxCoO3+δ (TC ) 335 K).42 However, insulating true ferromagnets with high TC’s are very rare. The typical values of TC are 100 K in BiMnO3,30,43 ∼200 K in LaMnO3+δ (δ ≈ 0.10),17,18 ∼200 K in La1-xSrxMnO3 (x ≈ 0.1),33 180 K (34) Moodera, J. S.; Santos, T. S.; Nagahama, T. J. Phys.: Condens. Matter 2007, 19, 165202. (35) Matsumoto, Y.; Murakami, M.; Shono, T.; Hasegawa, T.; Fukumura, T.; Kawasaki, M.; Ahmet, P.; Chikyow, T.; Koshihara, S.; Koinuma, H. Science 2001, 291, 854. (36) Macdonald, H.; Schiffer, P.; Samarth, N. Nat. Mater. 2005, 4, 195. (37) Serrate, D.; De Teresa, J. M.; Ibarra, M. R. J. Phys.: Condens. Matter 2007, 19, 023201. (38) Kittel, C. Introduction to Solid State Physics; Wiley: New York, 1996. (39) Mahadevan, P.; Aryasetiawan, F.; Janotti, A.; Sasaki, T. Phys. ReV. B 2009, 80, 035106. (40) Torrance, J. B.; Shafer, M. W.; McGuire, T. R. Phys. ReV. Lett. 1972, 29, 1168. (41) Shimakawa, Y. Inorg. Chem. 2008, 47, 8562.

12432

J. AM. CHEM. SOC.

9

VOL. 132, NO. 35, 2010

in K2Cr8O16,44 and 33 K in CrBr3.38 Before now, the highest Curie temperature for ferromagnetic insulators/semiconductors was reported for La2MnNiO6 (TC ≈ 280 K).24,45 Bi3Mn3O11.6 shows a new record high TC among insulating/semiconducting true ferromagnets. Insulating/semiconducting ferromagnets are highly attractive materials for spin electronics.34,46 In conclusion, we showed that the oxygen content in Bi3Mn3O11+δ varies in a wide δ range, keeping the same cubic structure and semiconducting properties of the material. We demonstrated for the first time that the oxygen content can vary for the same cation composition in KSbO3-type materials, and the oxygen content can be increased further up to BiMnO3.867 (Bi3Mn3O11.6), meaning that there is additional space in the channels of the structure. We also discovered true ferromagnetism in a semiconducting material, which is rather rare.24,34,38 Bi3Mn3O11.6 is a ferromagnet with the highest Curie temperature (TC ) 307 K) for this class of materials (ferromagnetic insulators/semiconductors). The Curie temperature of Bi3Mn3O11.6 is above room temperature, and which makes it very important for practical applications. Acknowledgment. This work was supported by World Premier International Research Center Initiative (WPI Initiative, MEXT, Japan), the NIMS Individual-Type Competitive Research Grant, and the Japan Society for the Promotion of Science (JSPS) through its “Funding Program for World-Leading Innovative R&D on Science and Technology” (FIRST) Program. We thank Mr. K. Kosuda of NIMS for electron probe microanalysis. The synchrotron radiation experiments were performed at the SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (Proposal Numbers 2009A1136 and 2010A1215). We thank Dr. J. Kim and Dr. N. Tsuji for their assistance at SPring-8. Supporting Information Available: Details of magnetization, resistivity, laboratory and synchrotron XRD, TGA, and dielectric measurements and bond lengths (PDF); X-ray crystallographic data (CIF). This material is available free of charge via the Internet at http://pubs.acs.org. JA1043598 (42) (a) Kobayashi, W.; Ishiwata, S.; Terasaki, I.; Takano, M.; Grigoraviciute, I.; Yamauchi, H.; Karppinen, M. Phys. ReV. B 2005, 72, 104408. (b) Sheptyakov, D. V.; Pomjakushin, V. Yu.; Drozhzhin, O. A.; Istomin, S. Ya.; Antipov, E. V.; Bobrikov, I. A.; Balagurov, A. M. Phys. ReV. B 2009, 80, 024409. (c) Fukushima, S.; Sato, T.; Akahoshi, D.; Kuwahara, H. J. Phys. Soc. Jpn. 2009, 78, 064706. (43) Moreira dos Santos, A.; Cheetham, A. K.; Atou, T.; Syono, Y.; Yamaguchi, Y.; Ohoyama, K.; Chiba, H.; Rao, C. N. R. Phys. ReV. B 2002, 66, 064425. (44) Hasegawa, K.; Isobe, M.; Yamauchi, T.; Ueda, H.; Yamaura, J.-I.; Gotou, H.; Yagi, T.; Sato, H.; Ueda, Y. Phys. ReV. Lett. 2009, 103, 146403. (45) Kang, J.-S.; Lee, H. J.; Kim, D. H.; Kolesnik, S.; Dabrowski, B.; Swierczek, K.; Lee, J.; Kim, B.; Min, B. I. Phys. ReV. B 2009, 80, 045115. (46) Gajek, M.; Bibes, M.; Barthelemy, A.; Bouzehouane, K.; Fusil, S.; Varela, M.; Fontcuberta, J.; Fert, A. Phys. ReV. B 2005, 72, 020406.